Has facial recognition identified Anne Boleyn in a Hans Holbein drawing? Anne Boleyn’s appearance

Added: 2026-05-12

[00:00:00]Has facial recognition software identified a Hans Holine sketch as the true face of Anne Berlin? That's what a new peer-reviewed article is asserting, along with some astounding claims about possible pictures of Anne's mother and sister and a pretty radical theory about who is really shown in a famous image of a blonde woman labeled as Anne, which I think will make your jaw drop and which I haven't seen getting much press attention. Can we trust this research though, or is this another crackpot theory about a picture of an like some of the others I've reviewed on this channel? Stick around to hear my thoughts.

[00:00:48]Okay. The new theory appears in a peer-reviewed article called reassessing an Berlin and other Berlin Women in Holine drawings using facial recognition. It was written by Karen Lel Davies Hassan Ugil I'm sorry if I didn't pronounce that correctly and David G.

[00:01:03]Stork and was published in NPI Heritage Science. I've left it linked below for you. It's free to read and I do recommend you take a look at it because it's quite a scientific read and I'm a historian so it does no harm to double check that I've understood it correctly.

[00:01:18]From what I've read in the press, Miss Davies, who is an amateur historian, initiated this project. In fact, the way it came about actually reminds me a little bit of how the project that discovered Richard III's remains began.

[00:01:30]The BBC says that since August 2024, Davies has been working as a cleaner to fund her true passion, historical research. One day, when Davies was cleaning a client's house, she told them about her passion project. They put her in touch with Professor Ugil and the rest is history if you'll forgive the pun. Ugiel is a mathematician who works at the University of Bradford and who specializes according to his university profile in the areas of geometric design and visualization, computer-based physical analysis, design optimization and machine learning. His profile adds that he is well known for his work on computer-based human face analysis, including face recognition, face aging, emotion analysis, and lie detection. As for David G. Stor, he works at Stanford and has a background in art history and physics with expertise in areas including rigorous computer image analysis of fine art paintings and drawings. According to the article, Davies did the historical research.

[00:02:27]Ugale developed the facial recognition software and did the analysis and stork quote critically reviewed the methodology results implications and relation to other computer-based methods.

[00:02:39]So how did they come up with the idea that they've potentially identified a contemporary image of anal? Well, they started with a drawing of an unknown woman by Hans Holbine the Younger held by the Royal Collection Trust and known as RC912190.

[00:02:56]We know that this woman was drawn whilst Holbine was working at the court of Henry VIII. And we know that the drawing dates to between early 1532 when Holbine arrived in England for the second time and his death in 1543.

[00:03:09]This is because all his drawings from that trip are done on a kind of pinkish paper which distinguishes them from drawings done during an earlier visit in the 1520s.

[00:03:18]Anne was queen from 1533 until her execution in May 1536 when she was between around 28 and 35 depending on when you think she was born. I think she arrived in about 1501 but 1507 is another popular date. The unidentified sitter in the drawing which is a preparatory sketch for a painting which was either never completed or which has been lost to history is shown from just above the waist up. has brown hair and eyes and a slim build just as contemporary reports of Anne describe.

[00:03:49]She stares straight out at the viewer with a neutral expression and is wearing a distinctive wide-brimmed hat or hood which curves over the top of her head and which is very unlike the French and English hoods you typically see on tutor ladies, though it's not unique. More on that point later. Under this headdress is a cap which extends down so that it covers up her ears. There are decorations on her hat which are only very faintly rendered. And her dress has slightly puffed sleeves at the shoulders and a highbacked V- neckline which opens to show her white smoked up at her neck.

[00:04:21]She has arched eyebrows and a slightly square chin. She looks to be anywhere from her late teens to late 30s. It really is very hard to tell with drawings and paintings. Holine has annotated the drawing to note that her dress is samat, meaning velvet and dam mask. I'm just going to call this image unknown brunette lady from here on out as rcin 912190 is quite the mouthful.

[00:04:47]The idea that this is an is based firstly on historical evidence, namely that this image dates to the correct time period to be her. It matches contemporary descriptions of her as having dark hair and eyes, a slim physique. She famously mentioned her little neck shortly before her death, and the sitter is wearing rich clothing.

[00:05:05]I mostly agree with this though any court here getting their drawing done by Holine will have been able to afford nice clothes and I would say that an was also described as having a sworthy complexion which I don't see in this image and which is not mentioned in the article that's something which might be harder to render on paper though especially pink tone paper and which might be in the eye of the beholder so I'm prepared to let that slide you could argue that toned it down to match the beauty standards of the day which favor paleness however As we'll see in a minute, one of the cornerstones of this whole theory is that Holine's work is practically perfect and trustworthy. And saying he changed something and this isn't what the sitter truly looked like would undermine the entire project, though I happen to think it's a distinct possibility with any portrait artist.

[00:05:51]What are the other reasons given to suppose that this is an though?

[00:05:57]Part of the picture's known provenence and treatment are considered noteworthy by the authors. It was at one time owned by a man called Jonathan Richardson the Elder. And the authors argue that its precise silhouetting and professional remounting, a treatment shared with only two other Holbind drawings owned by Jonathan Richardson, suggest early recognition of unusual historical value despite the absence of an inherited identification.

[00:06:22]I don't think the images treatment suggests anything in particular about the identity of the sitter. However, we have no idea when or why it was mounted or silhouetted. The Royal Collection just guesses that the backing paper is from the 17th century and it was owned by several other men, Richard Meade, Walter Chatwin, and Benjamin Wei before getting to the Royal Collection before 1792. Plus, we don't know who had it after Holine and before Richardson.

[00:06:47]Also, I don't know what silhouetted means in this context, but I think maybe it's a reference to the fact that the drawing was cut down to this size at some stage. Thoughts below, please, if you come from an art background and know what that's referring to. The next piece of evidence used to provisionally identify her and the bit that got the press's attention was facial recognition software. Essentially, the researchers had the software analyze 81 whole buying proprietary sketches, some of which are of known sitters who we are certain either were or were not related to Anne, including her maternal first cousins Mary and Henry Howard, and 13 of which are unidentified. By the by the authors acknowledged that even some of the attributions we have aren't completely secure.

[00:07:31]They also used one portrait of Anne's daughter Elizabeth completed in about 1546 when Elizabeth was 12 or 13 and attributed to William Scrotz plus a later anonymously painted portrait of Anne Berlin now held in the National Portrait Gallery in London painted in the 1580s or later and presumed to be a copy of a lost original by an unknown artist. That gives us 83 images in total. On the basis of this group, they argued that our unknown brunette lady showed physical characteristics which indicate that she was a member of the Bolin Howard clan and are not incompatible with her being Elizabeth's mother. In other words, they're attempting to use familial resemblance, but with a computer detecting it instead of human eye. It's important to note, and I appreciated this, that they do not claim that this is a slam dunk, despite some of the more hysterical media attention on this story. Instead, they say that our unknown brunette warrants reconsideration as a potential likeness of Anne Bolin.

[00:08:31]So, what are the strengths and weaknesses of this approach? Well, the researchers spend some time explaining why the software should be considered reliable, saying that facial kinship verification systems trained on modern data sets achieve accuracy rates exceeding 80% for determining parent child and sibling relationships based purely on facial structure. Okay, that's a good result. But those modern systems are presumably trained on data sets comprising photographs and footage, and they still have a roughly 20% failure rate. This team had to train their system on drawings and one painting.

[00:09:08]They do talk about this problem saying that in Renaissance studies, convolutional neural networks can distinguish between portraits of the same individual and different sitters with remarkable accuracy when sufficient training data exists. Again, that sounds good, but it's discussing portraits, not drawings. And I wonder how much training data those other models had, and if that data was of known individuals. We've only got 83 images total to work with here, 13 of which aren't even provisionally identified. Again, the authors offer an argument as to why their research should nevertheless be trusted despite these limitations. And it goes back to what I said a minute ago that Holine was basically such a good artist that we can completely trust his images. They say Holine's preparatory sketches functioned as authoritative working likenesses from which all subsequent works derived with systematic evidence establishing their accuracy.

[00:10:04]Systematic physical evidence demonstrates wholebind sketches functioned as definitive records.

[00:10:10]They go on to argue that the fact that we can see how the drawings were put together, sometimes slightly amended after a second sitting, and that some, like the Thomas Mur drawing, have details like pricking to allow the drawing to be transferred or pounced onto a panel or quote stylus tracing over outlines for panel transfer. Proves that these preparatory sketches prioritized structural accuracy for workshop reproduction rather than aesthetic idealization.

[00:10:38]They finish this little section by saying, "By restricting our analysis to these functional working documents, we address concerns about artistic interpretation and idealization that affect finished painted portraits."

[00:10:53]Uh, I'm afraid I don't find this series of arguments very convincing, though.

[00:10:58]And I'm going to tell you why right after I ask that if you're enjoying this content and would like more from me, please give the video a like, hype, and comment so that YouTube knows you approve. And subscribe with the notification bell on to stay up to date with my uploads. It does no harm if you're already subscribed to just double check that notifications are still on, by the way, as I get a lot of people telling me that they seem to randomly deactivate sometimes. For bonus material, you can also become a YouTube member using the join button under videos, or join me on Patreon, where I release things like patronon podcasts each week, early access to ad free videos, and recommendations for other great history content on the web. I'll leave it linked in the description box and pin comment for you, along with a link to my website, historyallingofficial.com, where you can join my free newsletter and get two complimentary downloads right away. Do double check that the welcome email you'll get when you join doesn't go to a junk mail though because if it does, the newsletter probably will too. And so you'll need to add my email address to your contacts list. I had a subscriber tell me recently, by the way, that Americans can no longer access some BBC news articles, which I used to link to in the newsletter, without paying a subscription. So rest assured that I'll no longer be using such links. Okay. So why am I not convinced by the arguments about using the whole bind drawings for this facial recognition project?

[00:12:17]I don't agree that the fact that you can see how a drawing was transferred to a panel means that the original drawing is photo accurate. Now I love Hans Holbine's work. Like love it. I have a whole biography video about the guy which I'll link for you. I think he was exceptionally gifted and I think his images have a lifelike quality rarely seen in other portraiture of the time. I have no doubt that he could render a person's face very accurately if he chose to do so and had sufficient time with his sitter. And I think a lot of the images we have which he created are likely excellent representations of their sitters which we could use to recognize those people if we timeraveled back to the court of Henry VIII.

[00:12:56]However, there are other things beyond an artist's ability which can affect the accuracy of the final drawing or painting. The fact that, as this article acknowledges, Holine made amendments to Sir Nicholas Point's portrait, for instance, which quote shows reworking of details, including beard, eyebrows, and eyelashes after a second sitting, shows either that Holine did not always get things quite right on the first sitting, or that he made changes based on input from the sitter. What if he had only one quick sitting with unknown brunette lately? What if she asked him to make her nose smaller or her eyes bigger or some other change? I know people point to the fact that some of Holine's sitters look a bit unattractive, including Henry the Yates's third wife, Jane Seymour, who looks quite stern, but gang, maybe he didn't have much time with them to perfect his drawings. Maybe that image of Gian is the prettied up version of her. Maybe she just wasn't as vain as others and didn't ask for alterations. Or perhaps she was confident in her looks. She had just bagged a king after all. The point is that Holine was a businessman. If the customer asked him to make tweaks, he was going to do it. I don't think that quote artistic interpretation and idealization are limited only to final portraits. They can be an issue in drawings, too.

[00:14:12]As much as I love his images and respect his exceptional skills, Hobine was also a 16th century artist, not a 19th, 20th, or 21st century photographer. And there are inherent limits in drawings and paintings when it comes to accuracy. You can even see differences in pictures done of the same person at around the same time. The two drawings of Anne's cousin, Henry Howard, exhibit very slightly different chin lengths to my eye and different hair colors because the profile view was abandoned before the hair was fully worked up.

[00:14:44]Furthermore, much later in the article, the authors actually acknowledge themselves the limits of the medium in which Holine worked, saying, "Even with Holine's highly disciplined practice, differences in lighting, paper type, or drawing tempo can subtly alter facial geometry, producing small but measurable deviations in the embeddings. All that being said, their software did produce some definite hits, which I think are worth mentioning. Comparing drawings of men like William Waram, Thomas Mr, and Thomas Wyatt to the blonde and brunette Holbine drawings and to Elizabeth's portrait did not produce any familial clustering as you would expect because as far as we're aware, those people weren't close blood relations. And despite how faint the rendering of Mary Howard's faces in her drawing, it gives an 85.5% similarity to a drawing known to be of her full brother, Henry Howard. These are good signs, but bear in mind that the two images of Henry, quote unquote, only had an 84.1% similarity, and they are of the same person by the same artist in the same medium at pretty much the exact same time. The pose is just a little different. Yet, the software thought Henry looked more like an incomplete faint image of his sister than himself.

[00:16:01]To be fair, the authors do say that different poses introduce a similarity reduction. But nevertheless, this collection of results had me scratching my head a little, wondering if I just wasn't grasping the science of it all very well, and wishing we'd had a few more examples to prove how trustworthy the facial recognition software supposedly is. We have whole bind drawings of Henry VIIth, admittedly not from life, Henry VIII, Mary, and a baby Edward V 6th. Did they produce a familial cluster? What about running Edward's picture against his mother Jane Seymour's drawing? We've got the Mr family painting and some preparatory drawings for it as well from the 1520s.

[00:16:41]There are biologically related people there too, including Thomas Mr and his father. Did such images produce significant results? The article doesn't say. What about other issues with this research? Well, there are a few mainly to do with the small sample size of images used to compare our unknown brunette lady to and the fact that no contemporary image of Anne herself was used as a comparison. 83 images is just not that many, especially when there is no verified an in the mix to act as a control. We don't know who a bunch of the images show, and many of those which are identifiable, we know aren't related to her. While the ones that are are at best just her cousins, sharing a mere 12.5% of her DNA, maybe a smidge more on account of some mild inbreeding down the generations, but that's not going to make much of a difference. Surprisingly, there were things I thought the researchers could have done to mitigate these problems, but they didn't.

[00:17:43]There was no discussion of the description of Anne's bones left by the Victorians in 1877.

[00:17:49]Even if the authors wanted to discount this, saying that we can't be sure the remains were an, I think a couple of sentences acknowledging this description's existence would have been a worthwhile inclusion. There is also one contemporary image of Anne's face, which is a 1534 portrait medal of her wearing an English hood and a cross necklace, but it is damaged, particularly on the nose. The researchers therefore made the decision to discount it entirely, which I thought was a mistake. I understand if the computer couldn't scan and work with a damaged metal, but it was sketched in the 19th century without damage, indicating that perhaps the squashing of the nose had not occurred yet, and that drawing could have been used instead.

[00:18:31]There was also a modern reconstruction of it, but that was definitely done post damage, so I understand being reticent about using it despite how good I think it is. There were other omissions too which I thought were unwise given the posity of holine images of people known to have been related to Anne. We have later portraits of her created in the 1580s and presumably based on lost originals no housed in the National Portrait Gallery in London and Heber Castle in Kent. There is also a ring called the checker's ring which almost certainly shows Anne and her daughter and which has been dated to about 1575.

[00:19:07]As I've already said, despite being postumous and anonymously painted, the NPG portrait is referenced in this study and used as a point of comparison. The other two are not. Now, the identification of Anne in the ring is disputed, so I appreciate why the researchers maybe wanted to discount it, but the Hea Rose portrait, so-called because Anne holds a rose in it, is definitely meant to show her, as is the MPG. They both have labels on them and are practically the same age, so of equal value in my opinion. And I thought the Heaver option should surely have been considered if the NPG was thought worth including. I also like the Heaver painting because it does show that sworthy complexion I mentioned earlier.

[00:19:48]You might recall that I did a video recently looking at another piece of research carried out by the staff at Heaver, which suggested that the NPG portrait, whilst meant to represent Anne, actually used a face mask of her daughter Elizabeth. Although I had reservations about that theory, the book it is found in by Owen Emerson and Kate McAffrey is very good and looks at Anne's portraiture more generally and I'll leave it linked below for you as well as my video on that theory. The reason I mention it here is that these new researchers did compare the NPG portrait of an to the picture of Elizabeth the first in the red dress and got a very high similarity match of 84.4% better than the match of 76.9% between the unknown brunette lady and Elizabeth.

[00:20:30]In fact, they argue that the relationship between these three images creates a coherent triangulation. The drawing that accords with contemporary written descriptions also aligns with the image Elizabethan's recognized as an relate appropriately to Elizabeth I.

[00:20:48]What would happen if they compare the unknown brunette to the Hebrew rose image though, which the Elizabethans also recognized as an, but which I don't think resembles the NPG all that strongly. Seriously, if the sitters weren't wearing the same clothing and jewelry, would you really think they were the same person? I don't know that I would. We never find out how the unknown brunette stacks up against the Hea Rose. However, though, as I said, I don't think there's any reason to privilege the NPG over the Hea Rose as an image of an and I say that as someone whose personal favorite happens to be the NPG. Before anyone asks, I also don't know that these results really support Emerson and Maccaffry's idea that the NPG used an Elizabeth and face mask because even if it did, they argued that it employed a face mask of Elizabeth dating from much later in her life, not from her childhood. Still, it's interesting to compare these different pieces of research and see what you think about how they might interact with each other. So, do check out that other video of mine and the article analyzed in this video. The lack of images of an to compare unknown brunette lady to is not the only issue.

[00:21:55]There are contemporary images of her close and fairly close blood relations available which were not used either. To name just a few, we have a picture of her father from his memorial plate on his tomb, which shows the same pronounced pointed chin that we see in the Hea Rose portrait, a feature I've always thought looks like an inherited one, and one which suggests the general accuracy of both images, at least to me.

[00:22:20]We have a whole bind portrait of her maternal uncle, the Duke of Norfolk. We have a portrait of her nephew, Baron Hunston. We have portraits of numerous great nieces and nephews, including Let No Nolles, plus a disputed image of Anne's niece, Katherine Carry, though as it's disputed, I think it's fair enough not to include it. And we have other portraits of Elizabeth, yet none appear here. Now, the omission of Thomas Berlin, Norfolk, Hunsden, and the great nieces and nephews is not directly addressed in the article, but the authors do mention that cross artist comparisons typically suppress coefficients, i.e. similarities due to stylistic and technical variance.

[00:22:59]However, they do look at other artists work when it suits, namely the image of Elizabeth and the NPG portrait of Anne, which wasn't even done from life. And they even say that in Elizabeth's case, despite the unknown brunette lady being a chalk on paperbind drawing and Elizabeth being an oil on panel by Scrotz, presumably the two still achieve a 76.9% similarity rating, which suggests that they are definitely related. That all sounds great, but bear in mind that another Holine drawing, which we've already touched on briefly, and which I'll call unknown blonde lady, it is labeled as Anne Bolin, but neither I nor the authors think it's really her, scores almost the same when compared to Elizabeth, getting 75.4%.

[00:23:42]The way the authors try to explain this away is pretty mindbending. So, do stay till the end to hear about it.

[00:23:51]So, portraits of Elizabeth by other artists can in fact show a close similarity to unknown brunette woman.

[00:23:58]Yet, no others are used. Why? Because the authors argue later portraits of her were very stylized as she morphed into Gloriana, unaging and covered in white makeup. I agree that this is a problem with many images of her in later life.

[00:24:14]However, not all of them refuse to show her aging. There's Amarcus Gearart's younger portrait of her in the 1590s which definitely does not do so for instance. And in any case, even if you're determined to avoid the Gloriana years, why was the image of her in the family of Henry VIII portrait aged about 11 discarded? Why were the images of her early in her reign when she was an adult and closer in age to her mother when Anne died not used either? The National Gallery of Ireland has a nice one from circa 1560, for instance, and a quick Google will show you some others. I take no issue with them not using her famous coronation portrait, by the way, as it's a later copy from the early 17th century. But, as I've just demonstrated, there are other options. The authors even admit there are problems comparing what they think is a woman in her 20s or 30s to a 12 or 13year-old child, saying that comparisons involving the adolescent Elizabeth the First introduce juvenile adult pairings that may understate parent child resemblance.

[00:25:14]Expanding the data set to include multi-age representations would enable a more systematic treatment of such variation. Despite this admission though, no further images of Elizabeth are introduced. The way the architecture of a child's face changes as they enter adulthood and even beyond is a very real issue though and one we can see in modern examples. I bet a lot of us have experience of seeing a photograph of someone we went to school with much later in life and we say, "Wow, I would never have recognized them." And I see this issue in celebrities too sometimes.

[00:25:46]I recently saw some press for the new Malcolm in the Middle reboot, for instance, and I thought one of the roles had been recast because the actor, 40-year-old Justin Burfield, who I hadn't seen a picture of since the original show ended when he was 20, looked so different to me. There's absolutely nothing wrong with how he looks. He just continued to grow and age, and I thought his face changed quite a bit. And that was comparing pictures of the same person who was an adult in both cases, not drawings and paintings of different people.

[00:26:13]I feel that it really hampers this kind of research to use only one picture of Elizabeth and one which shows her at such a different age to what Anne would have to be if this picture of the brunette lady actually is her. We also get a very debatable reading of the evidence when it comes to how accurate this picture of the future Queen Elizabeth is. The authors say that it is the most reliable pre-conographic likeness of Elizabeth painted when she was still a minor without significant political status. In 1546, Elizabeth had been recently returned to the line of succession as third in line to the English throne. She had serious political status and a portrait of her looking her best, which could be shown to ambassadors to encourage them to recommend her as a wife to their masters or master's family, was well worth having. Just because she wasn't Gloriana yet doesn't mean her images didn't have propaganda purposes. Nevertheless, this is the only one they use. All the while admitting that additional problems are caused by comparing across artists and media. I also thought far too much responsibility was put on the image of Elizabeth used. It is described as an independent control within the Howard family network being the first cousin once removed of Mary and Henry Howard. I raised an eyebrow at this as I struggle to see how someone's image functions as an independent control for their first cousins once removed. Unless we're looking at some serious inbreeding, which we aren't. I have plenty of first cousins once removed. And apart from superficial things like skin and sometimes hair or eye color, I don't look like any of them in particular, which is little wonder as we only share 6.25% of our DNA. I also just found it very unconvincing to try to identify brunette lady based partly on a supposed resemblance to just a couple of Anne's cousins with whom she only shared 12 and a half% of her DNA. Remember, especially as kinship networks were so vast in this era that there were a lot of people who were her cousins or her cousins cousins and so a lot of other women who might feature in this drawing. All that being said though, I will admit that brunette lady bears a resemblance to Mary Howard right down to the unusual hat. But I wondered if this was not because they were cousins, but because they are two images of the same person. Author Kate Herd suspects the same. In her book, Holbine at the Trutor Court, she writes that a second drawing in the Royal Collection, our unknown brunette lady, may be a companion study to Mary Howard, worked up further to portray the sitter's features. Certainly, if I had to pick someone to identify brunette woman as, I'd pick Mary. Her brother Henry had his drawing done twice by Holine, so why not her as well?

[00:28:56]Furthermore, this would explain why she's wearing a similar outfit to Henry if they were deliberately twinning for companion portraits. Plus, the 80.6% similarity result Mary Howard and Unknown Brunette got from the facial recognition software. a result which was one of the strongest in the study but which is buried in a table on page 10 of the article and not discussed in the main text perhaps because it is better than the comparisons between brunette lady and the NPG portrait of an and brunette lady and Elizabeth and might have undermined the whole project's premise that brunette lady is most likely an I'm not saying she's Mary Howard by the way just that Mary would be my choice long before I go for an and I'd argue that any similarity between the two images and indeed between brunette lady and Henry could also be explained away in this manner.

[00:29:48]I also just fundamentally disagree with the whole concept that familial similarity is trustworthy enough to identify unknown images from as there are so many people who don't look particularly like one parent or the other or even like their siblings. I don't think Prince George and Prince Louie look especially alike, for instance, but I have no doubt that they share 50% of their DNA. Sometimes you get people who look extraordinarily like their close family. Google Donald and Kefir Southerntherland, Reese Witherspoon and her daughter Eva, or Jennifer Garner and her daughter Violet for examples. And sometimes people don't. All of those kids don't look screamingly like their other parents, for instance, because they bear such a resemblance to just one parent. It's just a genetic lottery as to which bits of DNA you get from which parent and how this is expressed in your looks. You can get kids who look a lot like an uncle and or grandparent, but not particularly like their parents. You can get kids who just look like themselves and not very much like anyone else in their family.

[00:30:48]And no, it's not because they're adopted or the result of an affair or IVF with donor material. Their DNA has just combined in such a way that they don't look like some or all of their biological family all that much. Now, you may well recall way back at the start of this video that I promised you a discussion of some theories about pictures of Anne Bolin's mother and sister. So, let's look at those now.

[00:31:11]Remember the unknown blonde lady labeled as Anne Bolin? The authors don't believe this is an primarily because of the blonde hair. And I wholeheartedly agree.

[00:31:20]Though some argue that this was originally dark and the chalk rubbed off to reveal the undercoat, I think this is unsupportable as you're saying that the chalk came off in a perfectly uniform manner only on the hair without smudging anything on the rest of the image and whilst leaving the hairline sharp.

[00:31:37]You're also saying that for some reason this only happened on this one image of a brown-haired person, though there are many more such people in Holine's drawing collection. As for the claim that you can see dark brown chalk at the top of the hairline, that looks like a tiny bit of shadowing to me, nothing more. Tying this image to Anne is also problematic because of the sitter's heavier physique and night, very informal for a shooter queen, but I could potentially get over those if not for the blonde hair. As for the label, it was added hundreds of years later and has serious provenence issues, which are discussed in detail in my video on the blonde lady drawing. The authors also note that Wes Slouse Holler made an engraving in 1649 based on another Holine drawing which is now held in the British Museum and said that it was Anne Berlin rather than turning to the unknown blonde woman suggesting she was not associated with Anne in that period.

[00:32:29]Whilst I don't disagree with these assessments to be honest they are neither here nor there for me.

[00:32:34]Hypothetically, even if Holler thought unknown blonde lady was an, maybe he just preferred the other image more, which by the way is no longer believed to show her. For all these reasons, the authors therefore discount unknown blonde lady as an yet their facial recognition software gave a 78.3% similarity score when comparing the blonde and brunette ladies and a 75.4% 4% match between blonde lady and Elizabeth I, which is barely lower than the 76.9% match between brunette lady and Elizabeth the authors try to get around this very tight clustering of results which suggests that all these images show closely related individuals by wait for it suggesting that the blonde lady is Anne Bolin's mother Lady Elizabeth Howard Berlin countress of Wiltshire and Orand and that the drawing was perhaps mistaken for her daughter, just like the image of Henry Howard was accidentally mislabeled with his father's name. Gang, at this point, my eyebrows disappeared right up into my hairline.

[00:33:41]Elizabeth Howard was born no later than 1489 because she was married no later than 1501. That's when her jointure was finalized and a girl had to be at least 12 to marry in this era. This would make her at least 43 years old in 1532 when Holbine returned to England. And honestly, she was probably quite a bit older. I've given you the most extreme dates possible here, but hardly any girls got married at 12, and the wedding probably took place a couple of years earlier, too, according to an biographer, Eric Ives. While the drawing could have been done as late as 1538 when Elizabeth died. In fact, the authors actually argue that the double chin the blonde woman has could have been the result of illness, noting that Lady Wiltshire was recorded as being diseased with the cough in April 1536.

[00:34:30]And they say that it's probably an image of her from that year or later, which would have made her at least 47. I simply do not find this feasible. The blonde lady does not look old enough to be Elizabeth Howard in the 1530s, but her age and the serious doubts it raises about this image being her are not discussed in the article. It also has to jump through hoops to explain away the Wyatt family heraldry on the reverse of this image, especially as the software found no meaningful similarity with the whole bind drawing of the poet Thomas Wyatt. The authors do this by saying that this heraldry indicates ownership or provenence, not the identity of the sitter. But I ask you, why would the Wyatt family want or get a random drawing of Lady Wiltshire? And why doodle or have Holine doodle their arms on the back of it to show ownership instead of writing the owner's name and maybe even the name of this sitter on it? It's also the only Holine drawing with work on the back of the paper, by the way. So drawing arms on paper like this was not some standard way of identifying the owner. It just all makes no sense to me. I think either the heraldry indicates this woman was a member of the Wyatt family or as the Royal Collection Trust speculates, the page was maybe just used as a bit of spare paper at some point.

[00:35:46]Some of you might be wondering, does the article not suggest that unknown brunette lady or unknown blonde lady are married and sister? Well, the authors can't claim that because their computer model said that a drawing labeled Lady Vu, which we have a corresponding painting for, though it's only a copy of the original Holine, quote, demonstrates biometric clustering within the Bolin Howard family network, whilst also showing a woman wearing vaguely similar clothing to a portrait now known as Mary Bolin, painted in the 1630s, which exists in multiple versions, and which is attributed to Remius Van Limput. The VU portrait was also once mislabeled as an English queen in the 1680s. And on the basis of this, they think Lady Vu is actually Mary with that English queen label being an error made because she was the sister of a queen. As she looks nothing like unknown brunette or blonde ladies, they can't claim that one of those drawings might be Mary instead.

[00:36:43]Now, this is once again very problematic. There is absolutely nothing else to dispute this woman's identification as Lady Vu. Furthermore, her husband Thomas was drawn twice by Holine. So, it makes total sense that his wife was drawn at the same time. And Kate Herd notes that on the basis of the paper used, this the Lady Vu image is probably a pair with the drawing of Sir Thomas. She adds that all the VU drawings probably date from 1535, shortly before Vu retired from court, since they are on the same stock of paper as that used for the portrait of Nicholas Bourbon, which can be securely dated. None of these issues are mentioned in the article, let alone a counterargument offered to explain them away. Nor is it explained that Mary Berlin was in disgrace in 1535, having recently married without permission and been banished from court. She probably wasn't getting sitdowns with the court painter. Maybe in the 1630s when Van Limput was painting the image now known as Mary Berlin, he took inspiration for her outfit from Lady Vu because he didn't have a real image of Mary Berlin to work from. But honestly, the similarities with the drawing are pedestrian enough that they might just be a coincidence. We're talking about a black dress and English hood. The later painting not done by Holine also shows man's sleeves and a brooch, but it's clearly a different brooch to that scene in the Van Leut image. The necklace is different, too. And Lady Vo never had the distinctive red band of material around her hood or neckline that Van Leut's painting shows. Plus, her hood is pinned on the opposite side. Oh, and the original drawing indicates light colored eyes, while the later painting after Holbine's work and the Mary Berlin image both show a woman with brown eyes. Last but not least, the sitters just look like totally different women to me. To finish up, yes, we are nearly finished.

[00:38:29]Well done if you're still with me. Go ahead and give yourself a gold star in the comments. I noted some errors in the article. Nothing critical, nothing which undermines their arguments, but I was a little surprised these issues weren't caught in peer review. But maybe it wasn't peer-reviewed by an historian. I don't know. Anyway, just for the record, the authors say that the whole bind drawing of James Butler is misidentified as the Earl of Orand, but it's not.

[00:38:52]That's the title Butler inherited in 1539. The British Museum identification number provided for the unknown lady, later engraved by Holler is wrong. The number given in the article actually corresponds to a random biblical drawing of the flight into Egypt. And in the same paragraph they say that the NPG 668 portrait of an quote was recorded in Elizabeth and inventories and referenced as Ambblein by contemporaries within reach of living memory. They cite Roy Strong's 1969 book Chitter and Jacabian portraits for this information which I have a copy of so I checked it. What Strong actually said about its provenence prior to the NPG getting it was that it was quote purchased in 1882 from the Reynolds galleries. previous history unknown. I would add that we now know too that the portrait dates to 1584 or later because of dendrochronology which told us that that was the year the tree which provided the panel it's painted on was failed. So what are my final thoughts here? Well, I think the idea of using facial analysis on old drawings and paintings is certainly intriguing and may well have merit and I enjoyed reading this article and agreed with many points in it especially regarding unknown blonde lady. However, I'm not terribly convinced by the results of this specific study. Unknown brunette lady may be Anne Bolin for all I know, but there's nothing in this study or elsewhere to strongly suggest an identification one way or the other.

[00:40:18]I'm equally sure that she could be any number of other brown-haired women in their late teens to late 30s at Henry VII's court in the 1530s or early 1540s.

[00:40:28]I think we're on very shaky ground trying to identify someone on the basis of whether a supposed image of them looks like one portrait of their daughter, a few pictures of a couple of their cousins, and a much later postumous portrait of them, presumably copied from a lost original by an unknown artist. Drawings and paintings are not photographs. There's too small a sample here for meaningful comparisons, in my opinion, despite the existence of other images of both Anne and some of her close family members. Though even if they've been included, I think it's still very debatable if there would have been enough for the algorithm to do its best work. And the results are erratic.

[00:41:03]They suggest connections between people known to not be closely bloodrelated, forcing the authors to come up with a series of highly implausible solutions in inverted commas, which require cherry-picking historical evidence to even attempt to make them stick. How many times did I note in this video that relevant historical information like Elizabeth the First's political importance in 1546, other portraits of her whilst still a young woman, Elizabeth Howard's age, or Mary Berlin's social status in 1535, which would have undercut the project's findings just conveniently weren't discussed or were presented in a very debatable manner.

[00:41:39]mind. This is in an article which says it's providing results which quote integrate historical material and computational analysis. I'm sorry to say that as entertaining and thoughtprovoking as the read was, it didn't feel like a very comprehensive integration to me. Do take a look at the article yourself to learn more about how some of these images did and didn't match up according to the facial recognition software. By the way, I simply couldn't take you through all the results here for reasons of time. I'm afraid I don't think we're really any further along in our quest for a contemporary image of Anne beyond the 1534 medal, but it was an interesting read. The authors did at least base their ideas on a scientific analysis, even if I think it has problems. They didn't just make stuff up, as some others have done in the past when talking about images of Anne. And as I said earlier, I also appreciated that they did not try to present this as a sure thing, but rather as a theory.

[00:42:31]Again, not everyone is as measured as that. Let me know in the comments if you think unknown brunette woman is an and for more about her, her family, and Hans Holbine, try one of these videos next or take a look at the selection I've put together for you in the description box, which is more comprehensive. Whatever you choose, please enjoy. And until next time, keep learning.